Meet the Nigel AI Advisor
by Austin Hill, Leader of the Software Product Management team at Emerson Test and Measurement
Test engineers are under intense pressure to improve productivity and team collaboration while product complexity grows at an accelerated rate. The promise of artificial intelligence (AI) to assist with some of the more mundane tasks has been on the horizon, but companies are understandably cautious about implementing black box tools that remove engineers from the loop, especially in critical industries such as healthcare and transportation.
Over the last two years, our team at Emerson Test and Measurement (formerly NI) has been developing an AI tool designed to help address some of the challenges engineers face. At our annual NI Connect conference in April, we announced the creation of a single intelligence called Nigel AI Advisor, which has been trained across the NI software suite and built on Emerson’s secure cloud network.
Unlike general-purpose AI assistants, we trained the Nigel AI Advisor on top of a cutting-edge large language model (LLM) with a wide variety of inputs specific to testing and validation within the NI software suite, thereby delivering expert results.
For example, we know one of the most time-consuming tasks for new LabVIEW developers is trying to understand and explain somebody else’s legacy code. One of the first uses we developed was to analyze and provide details about Virtual Instrument (VI) files, including a list of all controls and indicators along with brief descriptions. The Nigel AI Advisor can also provide suggestions for changes or additions tailored to your specific needs.
LabVIEW is a powerful tool, but we also recognize that it can be daunting to use due to its numerous features and options. Sometimes, engineers may have an idea of what they want a VI to do, but not necessarily know which functions to use for that purpose. Now, you can ask questions in plain language to quickly surface the correct tools across more than 700 functions, along with examples that will help you learn how to use those functions.
I’m very excited to see how the Nigel AI Advisor will evolve. Since it’s built on top of the current OpenAI LLM, it will grow in intelligence as that model advances and our own refinements improve. Additionally, we can port it to a new LLM as necessary, should a more advanced model be released in the future that offers improved performance.
In our July release, the Nigel AI advisor will be incorporated into most existing LabVIEW and TestStand licenses at no additional cost. We plan to release it for other NI software over time. What we don’t want is for AI to completely write test and validation code. AI should assist engineers, not replace them. Furthermore, we recognize that data privacy is of paramount importance to many individuals working on sensitive projects—your data needs to remain yours, which is why we’re developing the capability to deploy to on-premises models, as well as secure models used by government agencies.
Truthfully, we’re developing all sorts of new tools in the lab that I’d love to show the world. It’s an exciting time in our field, but we’re moving forward with the knowledge that test and validation can’t follow the same startup mantra of “move fast and break things.” AI is changing the way we approach our work, but it doesn’t change the purpose of our work. Our job is to make sure things work in the field every time. And so, our tools must have the same high standards we hold ourselves to.